Experimental Investigation on Forced Convective Heat Transfer in Additively Manufactured Cooling Channels

Sammanfattning

This thesis presents an experimental investigation of forced convective heat transferin additively manufactured (AM) cooling channels representative of Siemens Energyturbine components. A closed-loop water rig with Joule-heated stainless-steel test objects was reconstructed, redesigned and validated to measure global and local Nusseltnumbers and Darcy friction factors over a Reynolds number range of approximately5 ×102 ≲ Re ≲ 2×104 and Prandtl numbers 4 ≲ Pr ≲ 9. The primary focusis on the accurate evaluation of the global Nusselt number and its dependence onReynolds number, Prandtl number and AM-induced surface roughness. The new rig combines high-precision mass-flow, pressure and temperature measurements with an automated post-processing framework in Star-CCM+. A solid-onlyconjugate heat-transfer model with temperature-dependent thermal and electricalconductivities is driven by a JavaScript controller, which iteratively matches simulated wall temperatures and electrical power to the experimental data. This redundant use of electrical and thermal energy balances allows consistent extraction ofsegment-wise heat-transfer coefficients, local and global Nusselt numbers, and friction factors. The system is first validated using a smooth stainless steel referencetube, demonstrating good agreement with the Colebrook–White friction correlationand the Gnielinski Nusselt correlation within their reported uncertainty bands. Subsequently, a 2×2 mm AM stainless-steel square channel is tested at four Prandtlnumbers. The results show clear roughness-induced increases in both Darcy friction factor and Nusselt number in the transitional and turbulent regimes, while thelaminar regime remains close to smooth-pipe behaviour. Entrance-region (thermalnotch) effects are shown to be significant for short heated lengths and are carefullyaccounted for in the interpretation of the data. A comprehensive error and uncertainty analysis, including Taylor-series error propagation and sensitivity studies onpressure-tap and temperature-probe placement, yields a total relative uncertaintyin the averaged Nusselt number of about 6 % and in the Darcy friction factor ofabout 15 %. Overall, the work establishes a validated experimental and numerical framework for characterising rough-wall heat transfer in AM cooling channels andprovides a basis for future correlation development and design optimisation.